Life circumstances and adolescent mental health: Perceptions ...

Thesis for the degree of Doctor of Philosophy, Sundsvall 2010

LIFE CIRCUMSTANCES AND ADOLESCENT MENTAL HEALTH Perceptions, associations and a gender analysis

Evelina Landstedt

Supervisors: Katja Gillander Gådin Kenneth Asplund

Department of Health Sciences Mid Sweden University, SE-851 70 Sundsvall, Sweden

ISSN 1652-893X Mid Sweden University Doctoral Thesis 93 ISBN 978-91-86073-89-3

Akademisk avhandling som med tillstånd av Mittuniversitetet i Sundsvall framläggs till offentlig granskning för avläggande av filosofie doktorsexamen 17 december, 2010, klockan 10.30 i sal M108, Mittuniversitetet Sundsvall. Disputationen kommer att hållas på svenska.

LIFE CIRCUMSTANCES AND ADOLESCENT MENTAL HEALTH Perceptions, associations and a gender analysis

© Evelina Landstedt, 2010

Department of Health Sciences Mid Sweden University, SE-851 70 Sundsvall Sweden Telephone:

+46 (0)771-975 000

Printed by Kopieringen Mid Sweden University, Sundsvall, Sweden, 2010

i

To my brother Niklas

ii

ABSTRACT Evelina Landstedt (2010) Life circumstances and adolescent mental health: perceptions, associations and a gender analysis Department of Health Sciences, Mid Sweden University, Sundsvall, Sweden ISBN 978-91-86073-89-3; ISSN 1652-893X, Doctoral Thesis 93 Despite a well-documented gender pattern of adolescent mental health, public health research investigating possible influencing factors from a gender-theoretical approach is scarce. This study aimed to explore what factors and circumstances are related to adolescent mental health and to apply a gender analysis to the findings in order to improve the understanding of the relationships between life circumstances and the gendered patterning of mental health among young people. The study population was 16-19-year-old Swedish students and data was collected by means of focus groups (N=29) and self-administered questionnaires (N=1,663, 78.3% response rate) in school settings. Mental health problems were defined in a broad sense including the adolescents’ own understandings, perceived stress, psychological distress and deliberate self-harm. The mental health problems of perceived stress, psychological distress and deliberate self-harm were twice as common among girls as boys. The findings suggest that adolescent mental health is associated with the life circumstances of social relationships, demands and responsibility taking and experiences of violence and harassment. Supportive relationships with friends, family and teachers were found to be of importance to positive mental health, whereas poor social relationships, loneliness and lack of influence were associated with mental health problems. Perceived demands and responsibility taking regarding school work, relationships, future plans, appearance and financial issues were strongly related to mental health problems, particularly among girls regardless of social class. The results indicate that physical violence, sexual assault, bullying and sexual harassment are severe risk factors for mental health problems in young people. Boys and girls experienced different types of violence, and the victim-perpetrator relationships of physical violence differed. These diverging experiences appeared to influence the associations with mental health problems in boys and girls. A gender analysis provides the tools to gain knowledge about the ways that boys’ and girls’ lives are shaped by gender relations and constructions at different levels

iii

in society and how these life circumstances represent risk- or protective factors for mental health. For example, unequal power structures and the ways girls are expected to ‘do’ femininity likely influence their life circumstances in ways that place them at greater risk of mental health problems. Hegemonic constructions of masculinity and advantaged positions likely contribute to life circumstances that are positive for mental health but are also implying risk factors for poor mental health among boys, e.g., violence. It is also important to recognise how the intertwined cultural and structural aspects of gender and social class influence the lives and mental health of boys and girls. In conclusion, gendered and class-related mechanisms at the different levels in society influence the distribution of risk factors unevenly among boys and girls, which could be a possible explanation for the gender differences in reports of perceived stress, psychological distress and deliberate self-harm. The likelihood of gender and socioeconomic differences in mental health problems should be taken into account in prevention and health promotion strategies at all levels in society. A greater awareness about gender relations and the gendered social circumstances under which young people live is required. The school environment is an important arena with respect to prevention and health promotion. There is also a need for a joint action against violence and harassment at all levels in society. Implications do not only concern young people; social policy and legislation should focus on reducing gender and class inequalities in general. Key words: Stress; Psychological distress; Deliberate self-harm; Students; masculinity, femininity; social determinants; social relationships; demands; responsibility taking; violence and harassment; school.

iv

SAMMANFATTNING Svensk titel: Livsvillkor och ungdomars psykiska hälsa: uppfattningar, associationer och en genusanalys. Trots ett väldokumenterat genusmönster i ungdomars psykiska hälsa finns det en kunskapslucka i den folkhälsovetenskapliga forskningen avseende genusteoretiska analyser av sambanden mellan ungas livsvillkor och psykisk hälsa. Föreliggande studie syftade till att undersöka vilka faktorer och omständigheter som är relaterade till psykiska problem, samt att analysera fynden ur ett genusperspektiv för att fördjupa förståelsen av relationerna mellan ungas livsvillkor och genusmönster i psykiska hälsa. Studiepopulationen var gymnasielever i åldern 16-19 år. Studien genomfördes i skolmiljö och data insamlades genom fokusgrupper (N=29) och en enkätstudie (N=1,663, 78.3% svarsfrekvens). En bred definition av psykisk ohälsa tillämpades vilken representerades av ungdomarnas egen förståelse, samt de psykiska problemen upplevd stress, psykiska besvär samt självskadebeteende. Resultaten visade att stress, psykiska besvär och självskadebeteende var dubbelt så vanligt bland flickor som bland pojkar. Psykiska problem var relaterade till livsvillkoren sociala relationer, krav och ansvarstagande samt utsatthet för våld och trakasserier. Stödjande relationer med vänner, familj och lärare var av stor betydelse för psykisk hälsa medan dåliga relationer, ensamhet och brist på inflytande var relaterat till psykiska problem. Psykiska problem var starkt kopplade till erfarenheter av höga krav och ansvarstagande avseende skolarbete, relationer, framtidsplaner, utseende och ekonomi, i synnerhet bland flickor oavsett socioekonomisk bakgrund. Resultaten indikerar att olika former av våld och trakasserier är allvarliga riskfaktorer för psykiska problem och att flickors och pojkars skiljda erfarenheter av olika former av våld samt relationen till förövaren, kan vara relaterade till skillnader i psykiska problem. Genusanalysen av resultaten föreslår att flickors livsvillkor påverkas av ojämlika maktstrukturer och konstruktioner av femininitet och att dessa livsvillkor bidrar till en ökad risk för psykisk ohälsa bland flickor. Livsvillkor kopplade till manlig överordning och hegemoniska konstruktioner av maskulinitet influerar sannolikt pojkars psykiska hälsa positivt. Dessa villkor kan dock också innebära risk faktorer för psykiska problem, t.ex. i fråga om våld. Studien uppmärksammar även hur kulturella och strukturella aspekter av både genus och social klass kan påverka

v

livsvillkor och psykisk hälsa för pojkar och flickor. Studiens slutsats är att genusifierade och klassrelaterade mekanismer på olika nivåer i samhället bidrar till en skev fördelning av riskfaktorer för psykiska problem vilket kan vara en möjlig förklaring till skillnaderna mellan pojkar och flickor i fråga om upplevd stress, psykiska besvär och självskadebeteende. Genus- och socioekonomiska skillnader i psykiska problem bör tas i beaktande i preventivt och hälsofrämjande arbete på alla nivåer i samhället. Detsamma gäller för en ökad medvetenhet om hur ungas livsvillkor är relaterade till psykisk hälsa och hur dessa villkor är genus- och klassrelaterade. Studien uppmärksammar skolan som en viktig arena för preventivt och hälsofrämjande arbete samt att gemensamma insatser krävs på olika arenor för att motverka våld och trakasserier. Implikationer av studien omfattar även generella samhällspolitiska insatser för minskad ojämlikhet. Nyckelord: Stress; psykiska besvär; självskadebeteende; gymnasieelever; maskulinitet; femininitet; sociala determinanter; sociala relationer; krav; ansvarstagande; våld och trakasserier; skola.

vi

TABLE OF CONTENTS ABSTRACT.............................................................................................................. III SAMMANFATTNING ................................................................................................V LIST OF PAPERS .....................................................................................................X PREFACE ................................................................................................................. 1 BACKGROUND ........................................................................................................ 2 CENTRAL CONCEPTS IN THE THESIS .....................................................................................3 Adolescence ....................................................................................................................3 Mental health .................................................................................................................3 Life circumstances ..........................................................................................................4 Gender ............................................................................................................................4 Social class/Socioeconomic status .................................................................................4 MENTAL HEALTH PROBLEMS - PREVALENCE AND UNDERSTANDINGS ..................................5 Stress ..............................................................................................................................5 Depression, anxiety and psychological distress .............................................................5 Deliberate self-harm ......................................................................................................6 How young people understand mental health ................................................................6 THEORETICAL FRAMEWORK ................................................................................................7 A public health approach ...............................................................................................7 A gender theoretical approach .......................................................................................8 SOCIAL DETERMINANTS OF ADOLESCENT MENTAL HEALTH ..............................................11 Family and peers ..........................................................................................................11 School ...........................................................................................................................11 Structure .......................................................................................................................12 Socioeconomic patterning ...................................................................................................... 12 Culture and media – Body image ........................................................................................... 12 Violence and harassment ........................................................................................................ 13

PREVALENT HYPOTHESES OF GENDER DIFFERENCES IN MENTAL HEALTH..........................13 WHY THIS STUDY? ............................................................................................................14 THE STUDY – AIMS AND RESEARCH QUESTIONS.................................................................16 METHODS .............................................................................................................. 17 MULTI-METHOD APPROACH ..............................................................................................17 CONTEXT ..........................................................................................................................18 THE QUALITATIVE STUDY (PAPER I)..................................................................................19 Grounded Theory .........................................................................................................19

vii

Focus groups ................................................................................................................19 Participants ..................................................................................................................19 Procedure .....................................................................................................................20 Analysis ........................................................................................................................21 THE QUANTITATIVE STUDIES (PAPERS II, III, IV) ..............................................................21 Data collection and procedure .....................................................................................21 Participants ..................................................................................................................21 The questionnaire .........................................................................................................22 Measures ......................................................................................................................22 Dependent variables – mental health outcomes ...................................................................... 22 Stress and stressors............................................................................................................. 22 Psychological distress ........................................................................................................ 23 Deliberate self-harm (DSH) ............................................................................................... 24 Independent variables ............................................................................................................. 24 Control variables .................................................................................................................... 24

Analyses .......................................................................................................................25 ETHICAL CONSIDERATIONS ...............................................................................................27 The qualitative study ....................................................................................................27 The quantitative study ..................................................................................................27 FINDINGS ............................................................................................................... 28 HOW IS MENTAL HEALTH UNDERSTOOD BY YOUNG PEOPLE? (PAPER I) ............................28 WHAT IS THE PREVALENCE OF MENTAL HEALTH PROBLEMS AND HOW ARE MENTAL HEALTH PROBLEMS PATTERNED? (PAPERS II, III, IV)........................................................ 28 HOW ARE SOCIAL RELATIONSHIPS RELATED TO MENTAL HEALTH? (PAPERS I, II, IV) ..............................................................................................................31 HOW ARE EXPERIENCES OF DEMANDS AND RESPONSIBILITY RELATED TO MENTAL HEALTH? (PAPERS I, II, IV) ..............................................................................................................33 Demands and responsibility – achievements ................................................................33 Demands and responsibility – gender performance, social relationships and financial issues ............................................................................................................................34 HOW ARE EXPERIENCES OF VIOLENCE AND HARASSMENT RELATED TO MENTAL HEALTH? (PAPERS I, III, IV) .............................................................................................................35 Victim-perpetrator relationship of physical violence ...................................................38 DISCUSSION .......................................................................................................... 40 ON THE RESULTS ...............................................................................................................40 Mental health problems ................................................................................................40

viii

Understandings ....................................................................................................................... 40 Prevalence and patterning ....................................................................................................... 40

Factors and circumstances related to mental health problems ....................................41 Social relationships................................................................................................................. 41 Demands and responsibility-taking ........................................................................................ 43 Achievements ..................................................................................................................... 43 Doing girl – doing boy ....................................................................................................... 45 Financial issues .................................................................................................................. 46 Violence and harassment ........................................................................................................ 47 Joking ................................................................................................................................. 47 Gender patterning, associations and a gender analysis ....................................................... 47

Summary of the gender analysis ...................................................................................52 Critical reflections on applying a gender analysis .......................................................53 METHODOLOGICAL CONSIDERATIONS ...............................................................................53 Multi-methods ..............................................................................................................53 The qualitative study ....................................................................................................53 The quantitative studies ................................................................................................55 IMPLICATIONS AND FUTURE RESEARCH .............................................................................58 CONCLUSIONS ...................................................................................................... 61 ACKNOWLEDGEMENTS....................................................................................... 62 REFERENCES ........................................................................................................ 64 APPENDICIES ........................................................................................................ 78 APPENDIX 1. INTERVIEW GUIDE – AN EXAMPLE ................................................................78 APPENDIX 2. EXAMPLES OF QUESTIONS IN THE QUESTIONNAIRE .......................................79 APPENDIX 3. EXAMPLES OF RESPONSES ON SPECIFIC QUESTIONS ......................................81 APPENDIX 4. OVERVIEW OF INTERACTION EFFECTS ..........................................................82

ix

LIST OF PAPERS The thesis is based on the following papers, which are referred to in the text by their Roman numerals: I.

Landstedt E, Asplund K, Gillander Gådin K (2009) Understanding adolescent mental health: The influence of social processes, doing gender and gendered power relations. Sociology of Health and Illness. 31(7), 962-78.

II.

Landstedt E, Gillander Gådin K. Seventeen and stressed – do gender and class matter? Submitted manuscript.

III. Landstedt E, Gillander Gådin K. Experiences of violence and reported psychological distress in 17-year-old students: a gender analysis. Resubmitted manuscript. IV. Landstedt E, Gillander Gådin K. (2010) Deliberate self-harm and associated factors in 17-year-old Swedish students. Scandinavian Journal of Public Health. DOI number: 10.1177/1403494810382941. In press.

Published papers are reprinted with the permission of the copyright holders.

x

PREFACE ”Child and adolescent mental health is a necessary priority for the healthy development of societies.” (WHO 2010). This thesis investigates what factors and circumstances are related to adolescent mental health and how a gender analysis can contribute to a deeper understanding of the links between life circumstances and the mental health of boys and girls in late adolescence. This research is inspired by my strong beliefs in equality, change and young people’s rights to well-being as well as and an endless curiosity, which has made and is still making me ask ‘why?’ From an early age, I have had deeply rooted beliefs in equality. In addition to leading me into the social and political arenas, these beliefs have inspired me to pursue university studies in philosophy, sociology and gender studies. Later on, these social, political and theoretical paths merged into the study of public health sciences. I became more aware of inequalities in health and the social and gendered distribution of determinants of health. Around year 2001, alarming reports about the deterioration of young people’s mental health, particularly among girls, became increasingly frequent. Adolescent mental health was on the public health agenda, and in my view, young people’s rights to health and well-being were under threat. I believe in the rights of young people, and I strongly object to circumstances that jeopardise their rights and well-being, such as drugs, marginalisation, violence, inequality and mental ill health. I have not only been driven by a commitment to political and social issues, but I am also and have always been driven by a curiosity that I have had the chance to develop in academia. Through questions like ‚how is knowledge created and how does knowledge lead to change?‛, research became tremendously interesting and appealing. An undergraduate thesis in sociology gave me the opportunity to start exploring the field of adolescent mental health. The key insights I gained from this work include the tendency in research to acknowledge the persistent gender patterning of mental health but to not ask the question of why it exists or apply a critical gender analysis. These issues were further developed in an empirical master’s thesis in public health sciences and, currently, in a Ph.D. project. Not only is it social injustice in itself that a generation of young people report elevated levels of mental health problems, but the prevalent gender pattern raises gender-equality issues both in terms of the need to acknowledge the existing gap between boys and girls, but also and foremost to better understand what is creating this inequality. Such knowledge is needed in order to prevent mental ill-health, and more importantly, to promote good mental health. This thesis, I hope, is one way of making advances towards those goals. 1

BACKGROUND Adolescence represents a complex transition from childhood to adulthood, which inevitably implies new challenges for boys and girls. These challenges include aspects related to mental well-being that not only stem from changes in relationships and demands related to the emergence into adulthood in a globalised world, but also from existential thoughts and developmental changes (Rutter and Smith 1995; Zubrick et al. 2000). That is to say, one could neither expect nor strive for the total absence of mental distress in young people; it is a part of being young (Michaud and Fombonne 2005). Nevertheless, there are many reasons to be worried about the mental health of adolescents, of which some are outlined below. First and foremost, mental ill-health (including mental disorders) represents a major global health problem that affects young people (Kolip and Schmidt 1999; Patel et al. 2007). Apart from the individual suffering, the burden of mental illhealth can be quantified by the measure of disability-adjusted life years (DALY). According to a review of research on adolescent mental health, mental disorders in young people contribute to up to 70 percent of the total DALY (Patel et al. 2007). Poor mental health during adolescence is also a risk factor for mental illness in adulthood (Aalto-Setälä et al. 2002; Fergusson and Woodward 2002). In addition, over the past decades, there has been an increase in the rates of selfreported mental health problems among young people in Western countries, but there has been no comparable rise in those rates in older age groups (Rutter and Smith 1995). This trend has also been shown in a Nordic context (Berntsson and Köhler 2001; Hagquist 2009) as well as in other European countries (Collishaw et al. 2004; Fombonne 1998; Gunnell et al. 2000; West and Sweeting 2003). It has also been found that this increase is particularly prevalent among 15-year-old girls (Hagquist 2010, Hagquist 2009, Sweeting et al. 2009). Less is known about the trends among girls and boys in late adolescence. Although the trends in the occurrence of mental ill-health in young people are alarming and deserve attention, they are not the main focus of this thesis. Taken together, mental health problems represent a significant burden to both adolescents and societies and are public health issues of high priority (Kolip and Schmidt 1999; Patel et al. 2007; Sawyer et al. 2007; Zubrick et al. 2000).

2

Central concepts in the thesis Adolescence Adolescence is a fluid, culturally sensitive concept that can be defined as a phase in life between the ages of 10-19 years (WHO 2005). Others prefer using the notion of ‘young people’ when referring to individuals aged between 12 and 24 years (Rutter and Smith 1995). This thesis focuses on people of ages 16-19 years, which is a period described as ‘late adolescence’ and the individuals who are the subject of this thesis are referred to as ‘adolescents’, ‘students’ and ‘young people’. Mental health The field of research on adolescent mental health encompasses a wide range of disciplines, such as psychology, psychiatry, sociology, education and the public health sciences. Consequently, the list of definitions of mental health is endless, although it is likely that most scholars within these fields would agree that mental health refers to an individual’s emotional and psychological well-being as well as the presence or absence of a mental disorder. The World Health Organisation argues that mental health is more than the absence of a disorder and defines it as follows: ‚Mental health can be conceptualized as a state of well-being in which the individual realizes his or her own abilities, can cope with the normal stresses of life, can work productively and fruitfully, and is able to make a contribution to his or her community‛ (WHO 2007). In this thesis, a broad empirical conceptualisation of mental health, or more precisely, mental health problems, is asserted. Rather than targeting specific disorders (e.g., diagnosed depression or anxiety syndrome) several self-reported mental health problems are in focus: perceived stress, psychological distress (mainly depressive and anxiety-related problems) and deliberate self-harm. These problems are considered to represent different aspects of mental health, which although not investigated in detail here, are assumed to be interrelated. This interrelationship is exemplified in Figure 1, which also illustrates the three mental health problems on a continuum where perceived stress represents the least severe state and deliberate self-harm the most severe state.

Figure 1. Schematic illustration of the mental health problems examined and the relationships between them. 3

Life circumstances The concept of ‘life circumstances’ refers to the material as well as cultural and psychosocial conditions under which people live. Other terms used in this thesis are ‘contextual’ or ‘environmental’ factors. Life circumstances relate to the social determinants of health, which are considered in more detail and with examples in the section on the public health approach and the review of the literature on factors associated with adolescent mental health. Gender Gender refers to the cultural and social constructions of what it means to be a man/boy or a woman/girl in a given society and how that is enacted in social practice. Gender also refers to a fundamental organisational principle in society as well as social relations and hierarchies. The concept of gender is used throughout the thesis. However, in the statistical analysis, gender was operationalised as a binary category (boy/girl), and the adolescents were ascribed to each category based on their own indication in the questionnaire. Given that the focus groups were single sex and self selected, the participants chose their gender identity/position for themselves. All non-mixed groups were, however, constituted by biologically defined boys and girls. No transgender persons took part in the study. Because gender is the social structure that is central to this thesis, the concept is elaborated upon in more detail below. Social class/Socioeconomic status The terms social class and socioeconomic status (SES) are conceptualisations of the locations of people and groups of people within a social structure. There is a vast amount of theoretical and empirical work on the matter, for example, the Marxian, Weberian and Bourdieuan schools of thought, which are unfortunately beyond the scope of this thesis. Despite the risk of mixing theoretical understandings and definitions by using two concepts, social class and socioeconomic status (and in some occasions also socioeconomic position) both refer to the material (e.g., access to resources) as well as cultural and psychosocial aspects (e.g., education, values and ‘social capital’) of social stratification (Lynch and Kaplan 2000, Skeggs 1997). It is notable that most existing conceptualisations and measures regard adults and there is an ongoing discussion as to whether adolescents’ socioeconomic status can or should be defined and measured according to their parents’ class/SES (e.g., household income or education level) and to what extent young people’s own perception or indication of their socioeconomic status should be taken into account (Goodman et al. 2007; Hagquist 2007; West and Sweeting 2004). In the present thesis, this is of particular interest because the adolescents of interest (16-19-yearolds) are in the transition to adulthood. Details on the indicators of social class/SES used are found in the methods section. 4

Mental health problems - prevalence and understandings Stress Stress can be regarded as both an exposure (stimulus) and a response (outcome) (Ollfors and Andersson 2007). In the present thesis, perceived stress is seen as an outcome and a dimension of general mental health. Stressors, on the other hand, are considered sources of stress. As illustrated in Figure 1, it is likely that there is overlap between stress and psychological distress. Several studies suggest that stress is a mediator between, for example, an individual’s environment and mental health problems (Bovier et al. 2004; Hazel et al. 2008; Wagner and Compas 1990). There is also a large body of evidence showing associations between stress and various mental health problems (Adkins et al. 2008; Byrne et al. 2007; Compas et al. 1993; Hankin et al. 2007, McLaughlin and Hatzenbuehler 2009, Meadows et al. 2006; Rudolph 2002; Torsheim and Wold 2001). There is consistent evidence of girls reporting higher levels of general stress than boys (Gillander Gådin 2002; Rudolph 2002; Ystgaard 1997). Swedish data from the Health of School Children Study (HBSC) show that girls in grade nine (approximately 15-16-year-olds) report significantly higher levels of perceived stress than boys (Danielson 2006), a result also shown by a report on upper secondary school students from the Swedish Agency of Education (2007). Less scientific evidence is available on stress levels in older Swedish adolescents. However, Ollfors and Andersson (2007) concluded that, relative to boys, 16-19year-old girls reported higher levels of stress on the majority of the stressor investigated, for example stress related to school work, demands and the physical environment. Depression, anxiety and psychological distress The rate of mental health problems linked to depression and/or anxiety is approximately 15-25 percent in the general adolescent population in Western countries (Hankin et al. 1998; Patel et al. 2007, Sawyer et al. 2007, Zubrick et al. 2000). Overall, there is a substantial gender pattern where girls and young women are 1.5 to 3 times more likely than boys/young men to report depressive and anxiety symptoms (Aalto-Setälä et al. 2002; Ge et al. 1994; Hankin et al. 1998; Lewinsohn et al. 1998; Nolen-Hoeksema and Girgus 1994; Patel et al. 2007). With respect to a broader notion of self-reported psychological distress, a large body of evidence shows the same trends. For example, in a Scottish sample of 15-year-old students, psychological distress (measured by the General Health Questionnaire GHQ-12) was reported by 44.1 percent of the girls and 21.5 percent of the boys (Sweeting et al. 2009). The corresponding GHQ-rates among young adults in Sweden (18-24-year-olds) has been found to be 32.9 percent among young women and 17.3 percent among young men (Nilsson et al. 2010). Similarly, according to a 5

Swedish study, psychosomatic symptoms were nearly three times as common among 15-16-year-old girls (19.8 percent) as boys (7.4 percent) (Hagquist 2009). Results from the cross-national study Health Behaviour among School Children (HBSC) showed that approximately 25 percent of European and North American girls (15-year-olds), compared to 16 percent of boys, reported having symptoms of psychological distress several times a week (Torsheim et al. 2006). In addition to showing the salient gender differences in adolescent mental health, this short review highlights the lack of studies of young people in their late adolescence. Deliberate self-harm Deliberate self-harm has been referred to as an act with a non-fatal outcome in which an individual deliberately initiates specific behaviours (e.g., self-cutting) or ingests a substance, drug or object with the intention of causing self-harm (Hawton et al. 2002; Ystgaard et al. 2009). Some researchers also include in the definition that the self-injury is not life-threatening and is without suicidal intent (Laye-Gindhu and Schonert-Reichl 2005). Evidence from population- or community-based American, Australian and European studies in young people report a lifetime DSH prevalence of 7-15 percent (8-24 percent among girls and 4-10 percent among boys) (De Leo and Heller 2004; Hawton et al. 2002; Laye-Gindhu and Schonert-Reichl 2005; Madge et al. 2008; Nixon et al. 2008 ; Young et al. 2007; Zøllner and Jensen 2010). Overall, most research on DSH is conducted on in-patients (Fliege et al. 2009; Lowenstein 2005). To my knowledge, there is only one Swedish, community-based peer reviewed study in which 40.2 percent of 14-year-old students (girls: 47.6 percent, boys: 33.3 percent) reported a lifetime history of self-harm related behaviour (Bjärehed and Lundh 2008). The high prevalence found in this study might be due to their broad definition of self-harm behaviours, namely, at least one act of DSH out of nine suggested by the Deliberate Self-Harm Inventory. For instance, if a girl or a boy had indicated ‘yes’ on the item ‘Carving words, pictures, etc. into skin’, he or she was classified as a DSH case. How young people understand mental health According to qualitative research, young people understand mental health as an emotional experience (Johansson et al. 2007) and adherence to normality (Armstrong et al. 2000; Secker et al. 1999). Young people’s perceptions of mental health tend to be associated with negative feelings such as sadness, worry, depression, loneliness, anger, or fear (Armstrong et al. 2000). With regard to young peoples’ understandings of depression, a Canadian study indicated that depression was mainly perceived as a withdrawal from others (Hetherington and Stoppard 2002).

6

Theoretical framework A public health approach Within the field of psychology and psychiatry, mental ill-health has historically been conceptualised as individual problems for which the sources are to be found within or in close attachment to the individual (Horwitz and Scheid 1999). Such individual sources of mental ill-health are, for example, hormones or genetics, (Angold et al. 1998; Costello et al. 2006; Zubrick et al. 2000), psychological characteristics (personality traits, rumination style) (Nolen-Hoeksema et al. 1999), and co-morbidity (other types of mental health problems) (Costello et al. 2006; Fliege et al. 2009; Patel et al. 2007). However, the underlying assumption of this thesis is that the main sources of adolescent mental health are to be found in the life circumstances of young people. This does not mean that individual-oriented approaches are irrelevant. As noted by a range of authors, the most likely explanatory model includes a combination of biological, psychological and environmental factors (Nolen-Hoeksema and Girgus 1994; Patel et al. 2007; Piccinelli and Wilkinson 2000; Zubrick et al. 2000). Nevertheless, this thesis is situated in the fields of public health sciences and medical sociology. Within these disciplines, it is argued that inequalities in health arise from the structural and organisational aspects of social inequalities, for example, how the unequal distribution of power and resources affect peoples’ lives and influence their risk of poor health (Baum 2003; Horwitz and Scheid 1999; Marmot 2007; Pickett and Wilkinson 2009). Given this perspective, mental well-being is assumed to deteriorate with declining social status and poor life circumstances (Aneshensel et al. 1991; Brown and Harris 1978; Kemper 1991; Pickett and Wilkinson 2009). A central component of this approach is to acknowledge the health promotion potential in identifying factors and circumstances that are possible to change and can be subjected to political influence and other influences (Dahlgren and Whitehead 2006). Hence, a public health approach focuses on social determinants of health: factors at different levels of society that influence or may influence health positively or negatively (Dahlgren and Whitehead 2006; Marmot 2007). Although somewhat simplified, these levels of social determinants are illustrated in Figure 2. Dahlgren and Whitehead (2006) underline the interaction between the levels: ‚This model for describing health determinants emphasizes interactions: individual lifestyles are embedded in social norms and networks, and in living and working conditions, which in turn are related to the wider socioeconomic and cultural environment.‛ (Dahlgren and Whitehead 2006, p. 19).

7

Figure 2. Model of social determinants of health. The model is derived from Dahlgren and Whitehead (2006).

Although this model pedagogically illustrates the determinants of health, one central social structure is neglected, namely, that of gender relations. In the original Dahlgren and Whitehead model of the social determinants of health, ’sex’ is referred to as an example of the biological characteristics of individuals ‚that influence their health and that are largely fixed‛(Dahlgren and Whitehead 2006, p. 19-20). As will be outlined below, it is important to acknowledge more aspects than biological sex in relation to mental health, such as what it is means to be a man/boy or a woman/girl. Given the public health recognitions of the social and structural determinants of health, it is striking how questions of gender as a social structure and the experiences of being a boy or a girl have received so little attention in research on adolescent mental health. As shown in Figure 2, gender has been added to the model in order to highlight its relevance as a social determinant of health on various levels in society. A gender theoretical approach The theoretical framework employed is based on a social constructionist gender theory that recognises gender as a fundamental, although complex, organisational principle in society. As illustrated in Figure 2, gender deals with gendered aspects of social structure (e.g., distribution of power and resources), culture (e.g., language, ideologies, sports, fashion, media), organisations (e.g., work-life, institutions, schools) and personality (e.g., identification, behaviour, sexuality)(Connell 2009; Kimmel 2008). Importantly, in contrast to the biological 8

category of sex, gender is not something we have - it is something we do, or perform, in social practice (Butler 1990; West and Zimmerman 1987). Such social practice is guided by discourses of masculinities and femininities, and is shaped by and reshapes structures in society. Moreover, as gender is ‘performative’, boys and girls are not passively socialised into static sex roles; young people learn to ‘do’ gender and many find these practices joyful (Connell 2009; Paechter 2007). There are, however, numerous examples of the restraining effects of practices of gender. Despite the existence of multiple representations of femininity and masculinity, girls and boys relate to, and are encouraged to adopt, dominant constructions and norms in terms of gendered beliefs and behaviours. That is, they are encouraged to adopt what is considered in Western society to be an acceptable performance of male or female gender within a heterosexual norm (Paechter 2006). Peachter (2006) argues that the ‘dominant’ versions of femininity and masculinity are ideal types rather than examples of how real people act and live their lives. Such dominant ‘ideal types’ of gender have been conceptualised as ‘hegemonic’ masculinity (Connell 2005) and ‘emphasised’ femininity (Connell 1987) or ‘hyperfemininity’ (Paechter 2006). According to Kenway and Fitzclarence (1997), hegemonic versions of masculinity represent ‚dominant and dominating forms of masculinity, which claim the highest status and exercise the greatest influence and authority and which represent the standard-bearer of what it means to be a ‘real’ man or boy‛ (Kenway & Fitzclarence, 1997, pp. 119–120). ‘Doing boy/man’ according to hegemonic masculinity involves, for instance, heterosexuality, sporting prowess, being capable of violence, toughness, inhabiting or aiming for power positions, competitiveness, strength and risk-taking. Another conceptualisation is that ‘doing boy’ is centred on not ‘doing girl’ (Connell 2005, 2009). Although there is a range of ways in which girls construct and enact collective femininity and their individual femininities, there are femininities that are more highly valued in contemporary culture (Paechter 2006). It has been argued that such ‘emphasised femininity’ or ‘hyper femininity’ are represented by, for instance, compliance, passiveness, dependence, beauty, empathy, sexually attractiveness and nurturance (Connell 1987; Paechter 2007). Femininities, or ways of ‘doing girl/woman’, do not confer power in the same ways as constructions of masculinity (Connell 1987). Instead, most femininities are constructed as various negations of the masculine; ‚the practice masculinity becomes ‘what men and boys do’, and femininity, ‘the Other of that’‛ (Paechter 2006, p.254). This illustrates that gender is a social relation, that is, ways that people, groups and organisations are connected and divided. Enduring or widespread patterns among such social relations constitute organisational specific arrangements (gender 9

regimes) as well as the overarching structure, gender order (or patriarchy), of a society (Connell 2009; Walby 1990). There are also power relations embedded in how individuals relate to and practice dominant ideals of masculinity and femininity (Connell 2009). Although structural arrangements are dynamic, the gender order is characterised by the inequality of men and boys collectively possessing higher status, more resources and greater power than women and girls. Power relations are maintained via cultural discourses as well through overt acts of dominance, such as violence (Connell 1987; Walby 1990). Being a boy or a girl is thus constrained by the opportunities and resources available. With regard to gender as a social determinant of health, a recent WHO report stated: ‚Gender relations of power constitute the root causes of gender inequality and are among the most influential of the social determinants of health.‛(Sen and Östlin 2007, p. xii) Gender is not only about differences between men and women, but also about constructions and relations within the groups of boys and girls. Gender is intersecting with other relations and structures such as social class, ethnicity/race, age, disability, sexual orientations (Shields 2008; Walby 1990; Walkerdine et al. 2001). The intersectional approach employed in this thesis implies acknowledging the ways young people’s lives and experiences are influenced by the complex interplay of gender and social class, more precisely, how mental health problems can be responses to the ways the interplay of gender and class inequalities shapes young people’s lives.

10

Social determinants of adolescent mental health This brief literature review focuses on mental health correlates on three different levels: family and peer relationships, school and structure (society) (see Figure 2). Because violence cuts across all these levels, this dimension is presented under the structural level. The literature review also covers some suggested hypotheses of the root causes of gender differences in mental health. Family and peers An extensive body of literature suggests that peer and family relationships are both risk and protective factors with respect to mental health problems among adolescents. There is, for example, strong evidence of the protective contribution of stable and supportive relationships (Armstrong et al. 2000; Haraldsson et al. 2010; Johansson et al. 2007; Patel et al. 2007; Rudolph 2002). With respect to risk factors, relational problems in the family and with peers, as well as in romantic relationships, have been found to be related to increased levels of stress in young people, especially among girls (Byrne et al. 2007; Rudolph 2002; Wagner and Compas 1990; Ystgaard 1997). Similarly, loneliness and relational problems (e.g., peer and parental conflicts, lack of social support) have been identified as risk factors for depression, anxiety and emotional problems (Brage and Meredith 1994; Brolin Låftman and Östberg 2006; Hankin et al. 2007; Heinrich and Gullone 2006; Kapi et al. 2007a; Kraaij et al. 2003; Wisdom et al. 2007). Prior research has also indicated that young people who self-harm report a history of poor social relationships and boyfriend/girlfriend problems (Bjärehed and Lundh 2008; De Leo and Heller 2004; Fliege et al. 2009). School The mental health associations of school-related factors are mainly centred around two dimensions: the school context as well as the content of school work. Both dimensions are highly relevant because young people spend a great deal of their time in school and doing school work. With respect to the school context, prior research suggests that adolescent mental health is influenced by the academic context of the school classroom, the socioeconomic standing of the school and the psychosocial work environment (Gillander Gådin and Hammarström 2003; Goodman et al. 2003a; Saab and Klinger 2010; Torsheim and Wold 2001). Other contextual factors related to mental health are student influence, student-teacher relationships and safety (Ellonen et al. 2008; Konu and Lintonen 2006; Modin and Östberg 2009; Simovska 2004). In their studies on health in 9-15-year-old pupils, Gillander Gådin and Hammarström (2003) recognise schools as gendered institutions and argue that gender relations in school settings may influence children’s health. With regard to safety, there is 11

strong support for negative mental health consequences of various forms of harassment in school. These aspects will be outlined in the section on mental health correlates of violence. Studies on the mental health influence of the content of school work (e.g., exams, marks, pressure) tend to be centred on two main paths. First, there is evidence of academic ability and success as being associated with positive mental health (Kaplan and Maehr 1999). Secondly, numerous studies show that academic stressors as well as pressure and worries about academic performance are risk factors for stress, psychological symptoms as well as DSH (Byrne et al. 2007; Hjern et al. 2008; Mahadevan et al. 2010; Murberg and Bru 2004; West and Sweeting 2003). Structure As noted above, the social structure in focus in this thesis is mainly gender, and to some extent, social class/socioeconomic status. A review of gender patterns in mental health problems was provided in the previous section on mental health problems - prevalence and understandings.

Socioeconomic patterning The evidence on socioeconomic patterning is somewhat inconsistent. Many studies show associations between disadvantaged social status and mental health problems. For example, young people of low socioeconomic status report elevated rates of stress, depressive symptoms and psychological distress, as well as deliberate self-harm (Goodman et al. 2005; Hawton et al. 2001; West and Sweeting 2003; Wight et al. 2006; Young et al. 2007). At the same time, there is evidence of weak support for SES associations (West and Sweeting 2004), especially if social class is indicated by parental SES (Hagquist 2007). It appears that the subjective perception of social class/SES, as opposed to family income or parental education level, better predict mental health in young people (Goodman et al. 2007; Hagquist 2007).

Culture and media – Body image One aspect at the societal level is the strong influence of cultural and media messages on young peoples’ perception of themselves and others (Aubrey 2007). Such messages and the hegemonic ideals of bodily shapes, beauty, trends and attributes, inevitably shape how young people relate to their bodies (Bengs 2000), which in turn, affects mental health (Siegel et al. 1999; Wisdom et al. 2007). A negative body image has been identified as a strong risk factor for low self-esteem and depression in both girls and boys (Allgood-Merten et al. 1990; Siegel et al. 1999). Others suggest that girls are more dissatisfied with their looks and bodies

12

than are boys, which potentially causes more mental health problems among girls (Polce-Lynch et al. 2001; Tolman et al. 2006; West and Sweeting 2003; Wisdom et al. 2007).

Violence and harassment There is consistent evidence that experiencing bullying and sexual harassment is related to mental health outcomes such as depression, psychological distress and DSH (Abada et al. 2008; Gruber and Fineran 2008; Nansel et al. 2004; Portzky et al. 2008). Several researchers argue that girls respond to sexual harassment more negatively than boys in terms of mental health (Gillander Gådin and Hammarström 2005; Gruber and Fineran 2008). It is also well established that experiences of physical and sexual violence/abuse are risk factors for depression, psychological distress and DSH in young people (Ackard and Neumark-Sztainer 2003; Fredland et al. 2008; Hawton et al. 2002; Schraedley et al. 1999; Thompson et al. 2004). Sexual and physical abuse have been found to be particularly detrimental for mental health among girls and young women (Fergusson et al. 2002; Hand and Sanchez 2000; Romito and Grassi 2007; Sundaram et al. 2004), whereas others identified stronger associations between sexual abuse and mental health problems among boys than girls (Haavet et al. 2004; Schraedley et al. 1999). Studies on violence within romantic intimate partner relationships (dating violence), suggest strong associations with mental health problems, especially among girls (Banister et al. 2003; Glass et al. 2003; Hanson 2002; Molidor and Tolman 1998).

Prevalent hypotheses of gender differences in mental health Attempts to explain gender differences in mental health problems target various levels, including biological factors, psychological traits, psychosocial factors and structural circumstances (Nolen-Hoeksema and Girgus 1994; Piccinelli and Wilkinson 2000; Stoppard 2000). With respect to gender differences in depression, reviews of existing evidence show inconsistent or weak support for biological explanations (Nolen-Hoeksema and Girgus 1994; Piccinelli and Wilkinson 2000). According to Nolen-Hoekseema and Girgus (1994), gender differences in depression in adolescents may arise because girls, compared to boys, have more pre-existing risk factors for depression before adolescence (e.g., a ruminative coping style and low levels of aggression and dominance in peer-interactions), and these risk factors cause depression when they interact with gender-specific biological (e.g., dissatisfaction with bodily changes) and social challenges that emerge in adolescence (e.g., sexual abuse and restraining feminine sex role). According to the gender intensification theory, many gender differences in mental distress are due to girls’ experiences of pressure to conform their behaviour to

13

gender expectations when they reach puberty (Priess et al. 2009; Wichström 1999). Similarly, gender conflict theory postulates that gender roles may result in personal restriction, devaluation or the violation of the self or others, which in turn may have negative consequences on mental well-being (Watts and Borders 2005). Pollack (2006), for instance, argues that a ‘boy code’ (e.g., shaming of emotional expressions) in society and the ways boys are socialised have negative effects on their mental health (Pollack 2006).

Why this study? Despite a well-documented, consistent gender pattern in adolescent mental health and a growing body of knowledge on the mental health importance of life circumstances, there are several gaps in the research. Overall, the general picture that emerges from the research reviewed is that the individual-focused approach is pervasive, and there is a profound lack of studies employing a gender analysis. For example, with respect to peer and family relationships, more knowledge is needed on the character of such relationships in girls and boys from different social backgrounds. In addition, research applying a gender analysis (and not a sex-role approach) is scarce. As noted, several aspects of the school environment have been shown to influence adolescent mental health. However, there is a gap with respect to the mental health associations of gender and the meaning of academic demands and success, including responsibility-taking. More research is also needed on gender and the school context in late adolescence. With regard to the structural level, few studies have jointly explored social and gender patterns in adolescent mental health. Mendelson et al. (2008) argue that the interaction between several sources of social disadvantage may cause ‘double jeopardy’ in terms of mental health influence. Despite some initial strong evidence, more research is needed regarding how different types of violence are related to mental health in adolescent boys and girls as well as whether patterns in victimisation are related to the rates of mental health problems. In addition, little is known about mental health associations of the victim-perpetrator relationship. In Sweden, this omission is particularly apparent in the case of dating violence. In addition, research within this field is often lacking a gender perspective and most work has focused on younger adolescents or on young adults and not the age group included in the present study. The currently prevalent explanations of gender differences in mental health distress are characterised by a deterministic perspective of gender role socialisation and a failure to acknowledge structural power relations. Great emphasis appears to be placed on individualistic perspectives of gender regarding ‘style’, traits and sex roles. Ideas of socialisation into sex roles have been critiqued for being dualistically deterministic and for assuming young people to be passive victims of society (Connell 2009; Hammarström 2002). Sex role theory also fails to

14

grasp the relational and power aspects of gender, such as the structural factors that restrain or directly affect girls and boys differently (Connell 2009; Hammarström and Ripper 1999). To summarise, in order to better understand, predict and prevent mental health problems and promote positive mental health among young people, advanced knowledge is needed on the relationships between mental health and the circumstances under which young people live. In addition, a gender analysis may provide new understandings of the links between such life circumstances and mental health. Another reason for this study is the sparse knowledge about life circumstances and mental health in late adolescence.

15

The study – aims and research questions The aim of the study was twofold. First, it aimed to explore understandings and prevalence of mental health problems and to investigate what factors and circumstances are related to adolescent mental health. The specific research questions were as follows: 1. How is mental health understood by young people? 2. What is the prevalence and patterning of the mental health problems o Perceived stress o Psychological distress o Deliberate self-harm 3. How are social relationships related to mental health? 4. How are perceived demands and responsibility taking related to mental health? 5. How are experiences of violence and harassment related to mental health? Secondly, the study aimed to apply a gender analysis to the findings in order to improve the understanding of the relationships between life circumstances and the gendered patterning of mental health among young people.

Papers in the thesis I

II

III

IV

Q1

Q2

Aim 1

Q3

Q4

Q5

Aim 2

Discussion Figure 3. Overview of how the aims and research questions are covered in the included papers. Q = research question.

16

METHODS Multi-method approach The thesis comprises four papers based on two sets of data generated by both qualitative and quantitative methods. One part originates from a focus group study with 16-19-year-old upper secondary school students. The other set of data is based on a questionnaire study of 17-year-old upper-secondary school students. Hence, the data were generated by a multi-method approach (Cowman 1993). The use of several methods improves the possibilities to produce rich data from which new understandings and knowledge could be generated. The project started with a qualitative study (grounded theory; see details below) that aimed to study experiences and perceptions related to mental health that are difficult to illuminate and capture in a quantitative study. These qualitative results guided the paths taken for the other three quantitative studies by inspiring the construction of the questionnaire. The quantitative approach provided tools to investigate prevalence, distributions, correlations and associations.

Table 1. Overview of design and methods used in the thesis. Paper

Design

Participants

Data collection Focus group interviews

Method of analysis Constant comparative analysis

I

Qualitative Grounded theory Explorative

Upper secondary school students age 16-19 years, 29 focus groups, N=104

II

Cross-sectional Descriptive

Upper secondary school students, age 17 years. N=1663

Questionnaire

Chi-square test Factor analysis

III

Cross-sectional Descriptive Analytical

Upper secondary school students, age 17 years. N=1663

Questionnaire

Chi-square test Logistic regression Rasch Factor analysis

IV

Cross-sectional Descriptive Analytical

Upper secondary school students, age 17 years. N=1663

Questionnaire

Chi-square test Logistic regression

17

Context The research project was carried out in upper-secondary schools in the County of Västernorrland, Sweden. This region has seven municipalities in both urban and rural areas and has approximately 250,000 inhabitants. The two largest municipalities have 95,000 and 55,000 inhabitants, respectively. Historically, the region has had an industrial socio-economic base supported by the forest industry. Over the past few decades, however, it has shifted towards a more post-industrial service economy. Approximately 38 percent of young people (males 32 percent; females 44 percent) who complete upper secondary education begin university studies within three years (The County Administrative Board of Västernorrland 2009). The unemployment rate among young people (15-24 year-olds) in the region is 29.5 percent compared to the national average of 25 percent (Statistics Sweden 2009). The study was school based. By the time data collection was conducted, there were 18 upper-secondary schools in the region, ranging in size from approximately 100 to 1,500 students. Most schools were public, although the number of independent schools has been increasing, particularly in the largest municipality. However, by law, no tuition fees are allowed in independent schools. The Swedish upper secondary school education is three years, normally starting when students are at the age of 16. It is not mandatory, although 98 percent of those who finish compulsory school (year 9) start upper secondary school. Of those, approximately 25 percent disrupt their education (The Swedish National Agency for Education 2008). The school system is organised into 17 educational programmes of different orientations, which can be broadly classified into ‘higher education preparing’ programmes (academic) and ‘occupational training’ programmes (vocational). The academically oriented education includes programmes on the social sciences, natural sciences, economics, languages, information technologies and art. The vocationally oriented programmes focus on, for example, child and recreation; construction; electrical engineering; vehicle engineering; business and administration; handicrafts; hotel, restaurant and catering services; industry; media production and health care. The vocational programmes are strongly gender-segregated (The Swedish National Agency for Education 2008). Existing data indicate that young people with working-class backgrounds and low parental education level are overrepresented in vocational programmes, whereas those with middle-class background and high parental educational levels to a greater extent choose academic programmes (The Swedish National Agency for Education; Hagquist 2007).

18

The qualitative study (Paper I) Grounded Theory The qualitative study relies on principles of constructivist grounded theory (Charmaz 2005, 2006). This approach was chosen because of the explorative aim of the study. The suggested flexible guidelines, principles and practises for sampling, data collection and analyses (Charmaz 2006) were also suitable for the study. Generally, grounded theory is used as a method to construct theories grounded in data (Strauss and Corbin 1998). However, this study neither aimed for, nor claimed to, develop a comprehensive new theory. Instead, through the recognition of social processes, the focus was to elaborate upon existing theories and evidence within the field (Charmaz 2006). Charmaz (2006) argues that rather than ‘neutrally discovering’ features in the data, constructivist grounded theory acknowledges that the analyses are social constructions and that they are contextually and theoretically situated and emerge from the researcher’s interactions within the field and interpretations of the data. One aspect of the constructivist grounded theory approach is to be theoretically sensitive and acknowledge how power relations should be recognised in research, for not only the power relation between the researcher and the participants but also the potential hierarchies within the context where the study is undertaken (Charmaz 2005). Focus groups The choice of focus groups was motivated by the method’s potential to generate rich data and capture cultural norms and shared experiences in a social context (Kitzinger 1994; Morgan 1996). Hence, group discussions can help the participants to explore, share and clarify their views in their own vocabulary (Kitzinger 1995). This was particularly important because the focus of interest was the dominant discourses to which the adolescents relate regarding mental health and the life circumstances that they believe influence mental health. Furthermore, focus groups may generate a feeling of confidence among the participants and reduce the power asymmetry in relation to the researcher, especially given the age difference between the students and the researcher (Gillander Gådin 2002; Kitzinger 1994). Participants Participants were recruited from schools in six municipalities. In order to obtain broad variation in experiences, focus groups were recruited with the goal of obtaining a sample with maximum variation (Patton 2002). This was mainly obtained through a selection of first-, second- and third-year school-classes representing different educational programmes (academic, vocational, maledominated and female-dominated). All students in each school class approached were asked to participate and the groups were self-selected, as discussed by Kitzinger (1994). At first, six focus groups were recruited. After the initial analysis 19

of the data, a theoretical sampling (see Charmaz 2006) was used, and the recruitment of another three groups was carried out due to perceived gaps in the data or identified issues we intended to further explore. Additional groups were recruited from theoretical educational programs and female-dominated school classes. Seven of the nine groups were then interviewed a second time. Of the two groups that were not interviewed a second time, participants in one group declined further participation and the participants in the second group declined for practical reasons. In order to broaden the background of the participants and further enrich the data, another 13 focus groups were carried out, predominately in the more rural areas of the region. The focus groups comprised three to eight students. As described above, the total sample varied by age and socioeconomic and demographic characteristics. Specific personal data were not collected at the individual level. The choice of single-sex groups was based on two main arguments. First, as Morgan (1996) argues, homogenous groups are preferable because they facilitate a context for confident discussions. Second, it was assumed that the impact of asymmetric gender-based power relations would be less prevalent in single-sex groups than in mixed-gender groups, as previously suggested (Gillander Gådin 2002). The students were asked to form single-sex groups. There was no intention of excluding anyone who wanted to participate; therefore, four mixed groups were included according to the students’ requests. Twelve of the focus groups were conducted with male groups, 13 were conducted with female groups and four groups were gender mixed. Procedure The focus groups were conducted in the participants’ schools. They lasted 60-120 minutes and were tape-recorded and transcribed. In order to generate a common point of departure for the following discussion, the participants were asked to reflect upon what they thought about and associated with the concept of ‚mental health‛ *In Swedish: psykisk hälsa+. Following this, the question ‚what do you think is important for adolescent mental health?‛ was asked. The discussions were intended to be broad and centred on the topics raised by the participants. The interviewer was guided by different themes such as friends, school, family, future plans and relationships (see Appendix 1.). According to the principles of constructivist grounded theory (Charmaz 2006), the content of the discussions was adjusted as the study proceeded to some extent and new insights regarding the processes influencing mental health were gained. Nevertheless, the main structure of the discussions was consistent throughout the study.

20

Analysis The analysis was conducted in a stepwise procedure. Initially, the material was read through several times in order to obtain a comprehensive picture of what the data were saying. Line-by-line coding was then carried out to conceptualise ideas, codes and expressions for what the participants described. Identified patterns or similarities influenced the direction of the discussion in the forthcoming focus groups as well as the coding process. Consequently, a constant comparative method was developed early in the analysis process to facilitate simultaneous involvement in data collection and analysis (Charmaz 2006). Preliminary broad categories were constructed by selecting relevant codes using a process of focused coding, followed by a process in which the properties of the categories were specified. The relations between the categories (social interaction, performance and responsibility) were further scrutinised and specified by theoretical coding (Charmaz 2006). Through a deductive analysis, the categories were revised and confirmed against the data. Workshops were also held with student groups not included in the actual focus groups. These discussions shed new light on the results as they emphasised slightly different aspects such as the importance of supportive family relations and financial issues. However, the outcome of these workshops generally confirmed the results.

The quantitative studies (Papers II, III, IV) Data collection and procedure The data were collected by means of a self-administered questionnaire during school hours and included 17 upper secondary schools in the region. Data collection took place during two weeks in April 2007 (mid-term) and was carried out with the assistance of teachers. The questionnaires were personally distributed to the teachers and collected by the same person (E.L). Answering the questionnaire took approximately 20-30 minutes. Participants The study population consisted of all students enrolled in year two of upper secondary school across all municipalities in the region. In five out of seven municipalities, students in all classes were invited to participate. In the other two municipalities (the largest municipalities), 50 percent of the school classes were randomly sampled. This method was developed mainly for practical reasons (primarily time limits). The total sample was 2,123 students. In the analyses, the random sampling was accounted for by a weight variable to make it possible to generalise the findings to an estimated full population sample of second year students in the region (approximately 3,195). In total, 1,688 students (79.5 percent) completed the questionnaire. After excluding those with missing data on gender, 21

the sample on which the analyses were based was 1,663 students (78.3 percent). Fifty-one percent (n=826) of the respondents were girls, and 49 percent (n=837) were boys. Forty-five percent of the students (n=709) attended academic educational programmes and 55 percent attended vocational educational programmes (n=861). Ninety-three students (5.6 percent) did not indicate the educational programme. Of the 455 students who did not complete the questionnaire, 12 percent (n=55) were whole school classes under vocational training at work places. Other plausible reasons were absence due to illness or absenteeism. In some schoolclasses, the non-response rate was approximately 50 percent, which might be due to data collection being carried out during a lesson in which the school class was divided into two groups. According to the notes from teachers, only two individuals openly declared that they did not want to participate. The questionnaire The content of the questionnaire was inspired by the findings of the qualitative study, such as the sections on stress, violence and psycho-social school environment. The content was also inspired by other questionnaires such as the WHO Health Behaviour of School Children study (HBSC) (Currie et al. 2008), ‘Young in Värmland’ (Hagquist and Forsberg 2007), and Statistics Sweden surveys (Jonsson et al. 2001; Statistics Sweden 2010). For further details on the measures of mental health problems, see the section on measures. Questions were discussed in focus groups and can thus be considered validated by a ‚think-aloud method‛ (Presser et al. 2004). Data collection was carried out over a short period of time, which did not make a pilot study possible. In total, the questionnaire comprised 61 questions, of which many were instruments consisting of several items. Measures

Dependent variables – mental health outcomes Stress and stressors In order to assess the perceived level of stress, respondents were asked to indicate how often during the past three months he/she had felt stressed. This way of assessing stress is similar to the methods used in other studies (Byrne et al. 2007; Ollfors and Andersson 2007; Torsheim and Wold 2001), although most research on adolescent stress has used checklists of stressful events to assess levels of stress (e.g., Compas et al. 1987). Response alternatives were on a scale from ‘always’ to ‘never’. Subsequently, perceived stress was categorised into three levels: ‘Any level of stress’ (always; often; sometimes; seldom); ‘High level of stress’ (always or often) and ‘Very high level of stress’ (always). Those who indicated any level of stress were asked to rate to what degree they experienced various factors as 22

stressful: ‘school marks’, ‘demands on oneself’, ‘future plans’, ‘lack of money’, ‘taking responsibility for others’, ‘looks’ (to look and dress in a specific way), ‘relationships with friends’, ‘home situation’ and ‘leisure-time activities’. These stressors were categorised into ‘yes’ (always/often) and ‘no’ (sometimes, seldom, never). As a means of exploring the dimensions of stressors, principle component analysis was employed and produced two factors with eigenvalues of 3.50 and 1.05 that explained 50.54 percent of the variation. Varimax rotation showed that the major component comprised social and relational stressors (lack of money, responsibility taking for others, looks, friends, home, leisure time) and the second factor comprised stressors related to achievements (marks, own demands and future plans). Psychological distress Psychological distress was measured using an index of six items. Respondents were asked: ‚How often during the past three months have you felt: nervous, anxious/worried, depressed/low, irritable, worthless or resigned‛. The instrument has similarities with other measures of psychological distress: the HBSC symptom check list, the Psychosomatic problem scale, GHQ-12 and Kessler scale (Andrews and Slade 2001; Dao et al. 2006; Hagquist 2009; Sweeting et al. 2009; Torsheim et al. 2006). The response alternatives were: 0. Never; 1. Seldom; 2. Sometimes; 3. Often; and 4. Always. A summed score for the six items ranging from 0-24 was calculated for each respondent (the higher the score, the worse the psychological distress). Similar to other studies (Griffin et al. 2002; Hagquist 2007) quartiles were used to classify individuals into cases and non cases. That is, individuals who scored 12 or higher (upper quartile) were considered to be cases of ‘psychological distress’. This cut-off is similar to the one suggested for the Kessler-6 scale (Kessler et al. 2002). The reliability of the scale was estimated using the Cronbach alpha coefficients resulting in a value of 0.83, which exceeded 0.70 and was thus acceptable (Kline 2000). The psychometric properties of the scale were evaluated by factor analysis as well as by the Rasch latent trait analysis and were found to meet the requirements of unidimensionality. The dimensionality of the scale was evaluated by exploratory factor analysis using Kaiser's criterion (eigenvalue >1) and Principal Component analysis and estimated by the maximum-likelihood method. The six items all loaded on one component, which explained 54.93 percent of the variance. The loadings for each item were: ‘Low’: 0.83; ‘Worried/anxious’: 0.81, ‘Worthless’: 0.79; ‘Resigned’: 0.70; ‘Irritable’, 0.67; ‘Nervous’: 0.62. The Kaiser-Meyer-Oklin value was 0.83, which exceeded the recommended value of 0.6, and the Bartletts Test of Sphericity reached statistical significance. 23

Based on concurrently approximations of item difficulty and person ability, the Rasch model focuses on the operating characteristics of a latent trait (Bond and Fox 2007). With respect to the goodness of fit (i.e., whether the items form a unidimensional construct and all items work in the same direction) the Rasch Infit MSQ ranged from 0.77 (worried/anxious) to 1.26 (worthless), which indicated a good fit to the Rasch model (Bond and Fox 2007). Different item functioning analysis (DIF) showed no difference in the systematic patterning of responses between boys and girls (DIF contrast 0.11 – 0.37). The item location analysis, i.e., a comparison of the distribution of item difficulty (severity) and person measures (‘ability’) along the latent trait, indicated that the items seem to function well to discriminate on the upper part of the scale (worse psychological distress) whereas items did not tap the lower end of the scale. This result is perhaps not a weakness since the measure of psychological distress (the latent trait) was aimed to identify those with the most severe psychological distress. Deliberate self-harm (DSH) A lifetime experience of deliberate self-harm was indicated by affirmation of one or both of these questions: ‘Have you ever deliberately inflicted harm to yourself (e.g., cut or burn yourself)?’ and/or ‘Have you ever deliberately taken overdoses of medicine in order to harm yourself? The response alternatives were ‘no’; ‘yes, once’; ‘yes a few times’; ‘yes, several times’. Those indicating any ‘yes’ were categorised as having a lifetime experience of DSH. The measure employed in the present study is similar to that of Hawton et al. (2002) and Ystgaard et al. (2009), which enabled comparisons of DSH-prevalence between countries.

Independent variables The independent variables (i.e., potential risk factors) are summarised in Table 3.

Control variables Controlling for confounding variables is crucial in research on health in order to reduce the risk that the effect of the exposure on the outcome is mixed up by factors associated with both the exposure and the outcome, such as demographic variables (Rothman 2002). Selected demographic variables are presented in Table 2. Educational programme and parental employment status were used as control variables in Papers III, IV and in the additional analyses presented in Table 6 and 7. In Paper IV, foreign extraction and family structure (the family structure the participant is currently living with) were also included as control variables.

24

Table 2. Demographic control variables used in analyses. Variable Educational program

Questions/items ‘Which educational programme do you attend?’

Scale/Categories Academic Vocational Indication of social class/SES

Parental employment status

‘What do your parents do/work with?’

Employed (both parents having a full or part time work or running own business). Not employed (one or both parents not having a full-time or part time job or running own business). The nonemployed category comprised those who were unemployed, retired, on sick leave, on parental leave, students, housewives/men, other and do not know. Indicator of financial standard in the family and SES.

Foreign extraction

‘Where were your parents born?’

Both parents born in Sweden One or both parents were born outside of Sweden

Living situation

‘Who do you live with, most of the time?’

Two adults (both parents or parent with new partner) Single parent or other (one adult, alone, own partner, relative, other)

Analyses Statistical analyses were performed using SPSS version 17 and Winsteps for the Rasch analysis. Factor analysis and Rasch analysis were used to evaluate composite measures. Between group differences were tested using Pearson chisquared statistics. Logistic regressions were used to examine the associations between potential risk factors and the mental health outcomes (very high level of stress, psychological distress and deliberate self-harm). All independent variables were categorical. First, unadjusted odds ratios were calculated for each independent variable in univariate logistic regressions. Secondly, each independent variable was adjusted for demographic control variables. All logistic regressions were performed separately for boys and girls, and 95 percent confidence intervals were calculated. The alpha-level of p